On 8/19/18 11:51 PM, Yu Li wrote:
bq. One attendee asked where does hbase want to go? Is it storage or a db system? Need to draw a sharp line. Do what we are good at. Wish more people could respond to this question, maybe throwing a separate thread even to the @user list? Shall we have some discussion in our PMC list? *HBase is a mature project but are we too old to innovate and evolve? Where is our passion and ambition? I want to see the answer. We need to see the answer.*
I'd like to think that maturity and ability to innovate/evolve are orthogonal. Perhaps if they are related, maturity (interpreted as stability as a software project) forces us to innovate less quickly to *retain that stability*.
I think HBase fills a very good niche. There are more things that we could do which would bring more to our users without invalidating that kind of system HBase is.
I'd like to think the efforts we're getting started on with the WAL+Ratis work are a good example of this. We're taking something "new" and introducing it to HBase in a manner that (architecturally) doesn't outwardly change HBase.
It makes me wonder if we could do better around how to phrase/present such efforts (e.g. Java's JCP/JSR, Python's PEPs, or even Kafka's KIP). Having some more structure around what is architecturally acceptable and the types of questions we expect to have answered would be nice. Parallel to the kind of lines we draw to "define HBase".
